How can I interpolate matrix?

0 0 0 0 31.2117

0 0 0 0 0

0 0 0 0 0

0 0 0 0 0

33.4177 0 33.0571 0 32.2147

I have 5*5 matrix.

I want to interpolate zeros using nonzero values.

How can I make this one?

[Stephen23]

Do you only want to interpolate, or do you want to extrapolate as well?

(If you only want interpolation, the left hand upper triangular section will be NaN)

[HyoJae Lee]

Just interpolation.

Thanks for helping.

Answer 1

[Stephen23]

M = [0,0,0,0,31.2117;0,0,0,0,0;0,0,0,0,0;0,0,0,0,0;33.4177,0,33.0571,0,32.2147]

M = 5×5

0 0 0 0 31.2117 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 33.4177 0 33.0571 0 32.2147

[Xi,Yi] = find(M);
[Xo,Yo] = ndgrid(1:size(M,1),1:size(M,2));
A = griddata(Xi,Yi,nonzeros(M),Xo,Yo)

A = 5×5

NaN NaN NaN NaN 31.2117 NaN NaN NaN 31.7632 31.4624 NaN NaN 32.3147 32.1344 31.7132 NaN 32.8662 32.6859 32.3851 31.9640 33.4177 33.2374 33.0571 32.6359 32.2147

[KALYAN ACHARJYA]

@Stephen Cobeldick Sir, to what extent can we interpolate, is there any certain ways to measure the acceptibility of interpolated data points, because in this example available data points are very few only?

And yes you have shown in the best way, in the first section of the given data, there are no reference data points.

Thanks for your answer! This is learning for me tool. Appreciated!

[Stephen23]

"is there any certain ways to measure the acceptibility of interpolated data points..."

No, there is no certain way to measure the acceptability. The usefulness of interpolation (and extrapolation) depends on having sufficient prior knowledge and understanding about the underlying physical process being modelled:

https://en.wikipedia.org/wiki/All_models_are_wrong

Answer 2

[darova]

Use scatteredInterpolant